From a583624503c620fbea48bd5e197b4bd53d64fbb0 Mon Sep 17 00:00:00 2001
From: Alireza Rezaei <84126554+RezaeiAlireza@users.noreply.github.com>
Date: Sun, 8 Jun 2025 13:56:11 +0200
Subject: [PATCH] fixed DS_Store, FLAGS, lxml

Added FLAGS(sys.argv) to initialize the flags.
Added .DS_Store to .gitignore and excluded in pack_data.py
---
 .gitignore                                             | 3 +++
 markuplm/examples/fine_tuning/run_swde/pack_data.py    | 4 ++++
 markuplm/examples/fine_tuning/run_swde/prepare_data.py | 3 ++-
 3 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/.gitignore b/.gitignore
index 05353e4e7..b94441370 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,6 +6,9 @@ __pycache__/
 # C extensions
 *.so
 
+# Ignore Mac DS_Store
+.DS_Store
+
 # Distribution / packaging
 .Python
 build/
diff --git a/markuplm/examples/fine_tuning/run_swde/pack_data.py b/markuplm/examples/fine_tuning/run_swde/pack_data.py
index 66550d4ee..b0b89ccb0 100644
--- a/markuplm/examples/fine_tuning/run_swde/pack_data.py
+++ b/markuplm/examples/fine_tuning/run_swde/pack_data.py
@@ -46,6 +46,7 @@
 flags.DEFINE_integer("first_n_pages", -1,
                      "The cut-off number to shorten the number of pages.")
 
+FLAGS(sys.argv) # unknown flag error will be raised if the flags are not set correctly.
 
 def pack_swde_data(swde_path, pack_path, cut_off):
     """Packs the swde dataset to a single file.
@@ -77,6 +78,9 @@ def pack_swde_data(swde_path, pack_path, cut_off):
     print("Start loading data...")
     for v in vertical_to_websites_map:
         for w in os.listdir(os.path.join(swde_path, v)):
+            # if is .DS_Store continue, o.w. causes an error in mac
+            if ".DS_Store" in w:
+                continue
             page_count = 0
             filenames = os.listdir(os.path.join(swde_path, v, w))
             filenames.sort()
diff --git a/markuplm/examples/fine_tuning/run_swde/prepare_data.py b/markuplm/examples/fine_tuning/run_swde/prepare_data.py
index c73b8e830..17f3921d1 100644
--- a/markuplm/examples/fine_tuning/run_swde/prepare_data.py
+++ b/markuplm/examples/fine_tuning/run_swde/prepare_data.py
@@ -31,7 +31,7 @@
 from absl import flags
 import lxml
 from lxml import etree
-from lxml.html.clean import Cleaner
+from lxml_html_clean import Cleaner
 from tqdm import tqdm
 
 import constants
@@ -51,6 +51,7 @@
     "The path of the output file containing both the input sequences and "
     "output sequences of the sequence tagging version of swde dataset.")
 
+FLAGS(sys.argv) # unknown flag error will be raised if the flags are not set correctly.
 
 def clean_spaces(text):
     r"""Clean extra spaces in a string.